Papers by Valeria Krzhizhanovskaya
Article Classification with Graph Neural Networks and Multigraphs (2024.lrec-main)
Copied to clipboard
| Challenge: | Existing and newly published articles require complex and complex pipelines to classify them into context-specific label taxonomies. |
| Approach: | They propose to enrich Graph Neural Network pipelines with multi-graph representations that encode multiple signals of article relatedness as distinct edge types. |
| Outcome: | The proposed methods improve the performance of a variety of GNN models compared to default graphs. |